Page 2 of 2

Re: Three-Point Estimating - Useful or Not?

Posted: Wed 02 Mar 2011 5:50 pm
by satisfactionuk
Hi,
I have acquired a copy of @Risk for projects and used it for some time now and found it to be very useful.

I like the fact that you can very easily do a variety of simulations based on the data from two and three point estimates for each task, group of tasks and the project as a whole.

This tool fits nicely in with the concept of the critical chain method where buffering is either not used at all for individual tasks or just used to control the points where non-critical tasks meet or interact with critical tasks. Using the critical chain method one allocates the buffer for both time and cost at the end of the project and then monitor its usage throughout the project. The idea is that if the ratio of project performance exceeds buffer usage then the project is on track and if the ratio of buffer usage exceeds performance then the project manager knows that is potentual trouble ahead.

When it comes to project selection and acceptance, using @Risk triangular simulation provides the decision maker with a mean value gained from X amount of simulations (say 1000 or 5000) a pessimistic value and an optimistic value. As with any estimating model the devil is in the detail and therefore the tails. What is good about the @Risk method is that the project manager presents not a fixed cost or time estimate to the decision maker but a series of risk values on a sliding scale.

The project manager can with a great deal of confidence say to the decision maker we can almost definitely complete this project 100% of the time at this cost and at this time value then go on to say there is a 50:50 chance that we can come in at this cost and time value (the mean) and if everything goes well we can do it at this cost and this value (say the 40% cost and time values). Then being a smart project manager s/he can pass the mouse over to the decision manager and ask him or her to define the amount of risk that s/he is willing to accept (risk appetite) with regard to the project.

This exercise alone really drives home to the decision maker that the ultimate finishing time and cost is not a definite figure but is determined by a sliding scale of risk that is dependent upon the risk attitude of that person making the decision. If the decision maker decides to go for the mean time of the project to set the time/cost budget which is 50:50 chance of success and the project comes in at 70, 80 or 90 percent (normally claimed to be a failure). Then the project manager can quite rightfully claim that this was not unexpected considering the overall risk profile the s/he presented to the decision maker at the beginning of the project.

The good thing about analysing the project in this way is that each phase of the project can be analysed in the same way as outlined above. This can prove to be really useful when considering the direct effect of unplanned changes on that particular stage and then for the knock on effects on subsequent stages and eventually the project as a whole.

When the above is considered seriously by project managers who have fallen foul of the cut and slash executive who instinctively believes that all project managers regardless of their ethical beliefs artificially pad out their project to protect their own personal status or integrity. It makes sense to be able to say to the decision maker, if you want absolute certainty then move the slider on the simulation to the 100 percent mark. However, if you decide to take a risk then that risk is solely your responsibility. Of course if the project actually comes in later than or cost more than the 100 percent figures shown in the simulation then the project manager should be really be considering if s/he is in the right job.

It has been reported that a major German company will not approve any project with a simulation risk profile that is less than 80% using this profiling method.

I hope readers of this article find it useful and spur them on to look more closely at the benefits of the critical chain method in conjunction with simulation analysis.

Kind regards

Stephan Toth